Did you know? Computers have eyes, too.

While still an emerging technology, computer vision is shaking up industries in new, innovative ways. It aids adopters in anything from making movies to monitoring traffic. And at Luxoft, we are working on the development of modern and future technologies that utilize computer vision, notably in Automotive. Computer vision is the key technology that lets the car “see” and analyze the environment. It makes possibilities like advanced driver assistance systems (ADAS) – which demand flawless safety and reliability – possible.

We are excited to share our computer vision expertise with C++ developers at the HackCV hackathon in St. Petersburg, which takes place October 7-8. The aim of the hackathon, which focuses on computer vision, is to introduce participants to real situations in safety-critical embedded system development and create high-level software for pattern recognition.

The truth is, it’s important to understand there is a need for computer vision. Why is it so disruptive, and why is everybody seemingly implementing it?


What is computer vision?

Computer vision is the process of taking data from images or video footage and transforming them into new representations. The transformation might be converting a grayscale image from a color image, recognizing a pedestrian crossing the street by analyzing color and brightness, or producing high-quality video footage by removing camera shake. To assist in the creation of these new representations, the input data may contain useful contextual information such as “the camera is mounted in a car” or “the laser scanner detects an object within a 10-meter distance”.

Because humans are visual creatures, this process may appear to be a simple task. After all, how hard can it be to identify a pedestrian when they are standing right there in front of you?

The truth is, seeing an object is more complicated than it sounds. For humans, the brain divides the object’s light hitting the eye’s retina into many channels, each streaming different information back to the brain. In response, the brain’s visual attention system identifies important streams to analyze – depending on the reason for viewing the object – while suppressing the analysis of others.

This massive amount of feedback involves widespread associative inputs from muscle control sensors and the senses – sight, smell, touch, taste and hearing – allowing the brain to draw cross-associations from past experiences. These feedback loops return to previous processing stages – including streaming to the brain and the “hardware sensors”, AKA the eyes, which control lighting via the iris – essentially “tuning” the reception of light on the surface of the retina.

However, in a computer vision system, the computer receives a grid of numbers from a camera or a disk with a microcontroller unit – and that’s it.

Computer vision systems, unlike people, do not actually see real-world objects. They only “see” a grid with numbers corresponding to the brightness or saturation value of each analyzed pixel in a photo. A camera in the computer vision system also lacks associative thinking and pattern recognition – both human skills that help us identify objects. This is what makes the process of developing computer vision systems so complex. Converting a table with numbers into a deterministic representation or perception, such as a pedestrian crossing the road, is no simple feat. But thanks to various algorithms, it can be done.

Why should I consider computer vision?

Computer vision helps automate routine processes, cutting costs and maximizing efficiency. All engineers have to do is build trainable systems that process incoming information from real-time visual data streams. Soon enough, we’ll be able to equip machines with the same self-learning cognitive abilities humans possess, eliminating the need to train the computer to learn a number-based table with associated data. In combination with the ability for computers to do monotonous work quickly and cheaply, these systems will positively impact the development of businesses across industries. Here are just some of the many uses of computer vision:

• Face and fingerprint recognition

• Surveillance monitoring for intruders

• Analyzing highway traffic

• Monitoring pools for drowning victims

• For moviegoers, computer vision is used to merge computer-generated imagery (CGI) with live action footage – by tracking feature points in the source video to estimate the 3D camera motion and shape of the environment. Widely used in Hollywood, this technique requires the use of precise mapping to insert new elements between foreground and background elements.

• For machinery, computer vision speeds up quality assurance inspections by using stereo vision with specialized illumination to examine machine parts. This includes measuring the tolerances of aircraft wings, auto body parts or finding defects in steel castings by using X-ray vision.

• In healthcare, using computer vision allows practitioners to implement medical image analysis (such as noninvasive diagnosis, image-guided radiotherapy and treatment planning), or perform long-term studies on brain morphology, following subjects as they age.

Computer vision in Automotive

Specifically, computer vision plays a key role in reshaping the Automotive industry. It's associated with the goal of making the car safer, more convenient and more intelligent.

Initially, computer vision was introduced as a technology that improves advanced driver assistance systems (ADAS), which help the driver and keep them safe. It was prevalent in the lane assist system (which helps keep the vehicle in the desired road lane), road sign recognition, obstacle detection (such as spotting pedestrians), blind area monitoring, road boundary detection, active cruise control and parking assistance systems. Further development of ADAS led to the creation of autopilots, which can take control of the car in relatively simple situations (such as highway traffic). Consequently, the next step for computer vision will be using it to develop control systems for autonomous vehicles.


Thus, the future of the Automotive industry is dependent on computer vision. For users, this technology is important because it allows automakers to create simpler and safer cars, improving the user experience. Drivers don’t have to worry about accidentally hitting pedestrians or driving outside of their lane, since the car “sees” where you’re going and the environment around you.

For example, one of the key computer vision applications for Automotive is technology that recognizes road signs. When driving, it’s easy to get distracted and accidentally miss a sign. This is potentially dangerous – it could have been a sign that warns about a dangerously sharp turn or a speed limit decrease in a school zone. To mitigate this, computer vision allows the car to “see” the sign, recognize it and show it on the dashboard display for the driver to see.

Computer vision for your business – Luxoft can help

As you can see, computer vision is being widely applied across industries as a cutting edge technology. It also integrates well with other sectors, such as artificial intelligence, robotics and autonomous vehicles, allowing for seamless experiences. If you wish to outpace your competitors, it’s important to consider computer vision when figuring out your business needs.

With solid expertise in computer vision and AI, Luxoft can help. We have assisted clients across industries, anywhere from Healthcare to Automotive.

Specifically in Automotive, we understand the development of an embedded system that utilizes computer vision is a serious challenge. The software has to be reliable, robust, integrated and open to future updates. Our experts are aware of these needs and implement them proficiently. We also have extensive experience in LIDAR, radar, sonar, cloud and handling immense amounts of data.

To find out how we can help your business, please reach out to us here.

See you at the Computer Vision hackathon!